Horton

Re: [PoC] Federated Authn/z with OAUTHBEARER

Jacob Champion <jacob.champion@enterprisedb.com>

View thread
From: Jacob Champion <jacob.champion@enterprisedb.com>
To: Peter Eisentraut <peter@eisentraut.org>
Cc: Daniel Gustafsson <daniel@yesql.se>, Antonin Houska <ah@cybertec.at>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>
Date: 2025-01-13T23:21:31Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. meson: Fix install-quiet after clean

  2. oauth: Run Autoconf tests with correct compiler flags

  3. Link libpq with libdl if the platform needs that.

  4. Doc: correct spelling of meson switch.

  5. oauth: Correct SSL dependency for libpq-oauth.a

  6. oauth: Fix Autoconf build on macOS

  7. oauth: Move the builtin flow into a separate module

  8. Remove a stray "pgrminclude" annotation

  9. oauth: Simplify copy of PGoauthBearerRequest

  10. oauth: Improve validator docs on interruptibility

  11. oauth: Disallow synchronous DNS in libcurl

  12. oauth: Fix postcondition for set_timer on macOS

  13. oauth: Use IPv4-only issuer in oauth_validator tests

  14. Work around OAuth/EVFILT_TIMER quirk on NetBSD.

  15. oauth: Fix incorrect const markers in struct

  16. Add missing entry to oauth_validator test .gitignore

  17. cirrus: Temporarily fix libcurl link error

  18. Add support for OAUTHBEARER SASL mechanism

  19. libpq: Handle asynchronous actions during SASL

  20. require_auth: prepare for multiple SASL mechanisms

  21. Move PG_MAX_AUTH_TOKEN_LENGTH to libpq/auth.h

  22. Make SASL max message length configurable

  23. jsonapi: fully initialize dummy lexer

  24. common/jsonapi: support libpq as a client

  25. Remove fe_memutils from libpgcommon_shlib

  26. Revert ECPG's use of pnstrdup()

  27. Explicitly require password for SCRAM exchange

  28. Refactor SASL exchange to return tri-state status

Attachments

On Fri, Jan 10, 2025 at 5:27 AM Peter Eisentraut <peter@eisentraut.org> wrote:
> Assuming that uintptr_t is the right underlying type for SOCKET, that
> seems ok.

Best I can tell, SOCKET is a UINT_PTR (distinct from PUINT) so I think
that's correct.

> #ifdef WIN32
>
> might not work because WIN32 is defined by the PostgreSQL build system,
> not by the compiler (see src/include/port/win32.h).

Ah, thanks for catching that. Changed to _WIN32.

On Thu, Jan 9, 2025 at 2:44 PM Daniel Gustafsson <daniel@yesql.se> wrote:
>
> > On 9 Jan 2025, at 23:35, Jacob Champion <jacob.champion@enterprisedb.com> wrote:
> > I'm sure someone's going to complain at some point, but IMNSHO, the
> > fix for them is just to use the same formatting and capitalization as
> > the discovery document, and move on.
>
> Fair enough, I buy that. Maybe the above could be de-opinionated slightly and added as a comment to help others reading the code down the line?

Done!

On Thu, Jan 9, 2025 at 12:40 PM Daniel Gustafsson <daniel@yesql.se> wrote:
> Support for oauth seems to be missing from pg_hba_file_rules() which should be
> added in hbafuncs.c:get_hba_options().

Done, and added a basic test for it too.

--

v41 handles the feedback since v40, continues fleshing out
documentation, and splits out three prefactoring patches:
- 0001 moves PG_MAX_AUTH_TOKEN_LENGTH, as discussed upthread
- 0002 handles the non-OAuth-specific changes to require_auth (0005
now highlights the OAuth-specific pieces)
- 0003 adds SASL_ASYNC and its handling code

When I applied Daniel's async_auth_portion patch from earlier, custom
flows began double-freeing their socket descriptors. That made logical
sense but it wasn't immediately clear to me how to fix it, until I
realized that "async authentication state" and "SASL mechanism state"
are two different things with different lifetimes, and my previous
attempts conflated them. 0003 introduces a cleanup_async_auth()
callback, to explicitly free the altsock and its related supporting
allocations as soon as it's no longer needed. Not only does this solve
the double-free, it removes an extra layer of indirection from 0004
and neatly fixes a TODO where the Curl handles were sticking around
for the lifetime of the Postgres connection. Assertions have been
added to keep the new internal API consistent.

pqSocketCheck() was returning ready if it found buffered SSL data,
even if an altsock had been set. I separated the two paths more
completely in 0003.

The FreeBSD 13.3 image started failing to correctly resolve libcurl
package dependencies, leading to missing libssh2 symbols at runtime.
And 13.3 went EOL at the end of 2024 -- which is possibly related to
the breakage? -- so I seemingly cannot perform a `pkg update` to try
to fix things. I've added a hack around this in 0006 that can
hopefully be removed again when our Cirrus images transition to 14.2
[1].

Next email will discuss the architectural bug that Kashif found.

Thanks,
--Jacob

[1] https://github.com/anarazel/pg-vm-images/pull/109